Azul Eases the Pain

We’ve all said it about one technology or another – PCs, cell phones, PDAs – “It’s a great technology… when it works.” The scale may be different, but many IT executives mutter the same phrase when it comes to their large-scale virtual machine-based deployments. As Java™, J2EE™ and .Net™ become the platforms of choice for vertical application development, the data centers supporting these applications are discovering that along with great new technology comes a good deal of pain.

According to Gartner, more than 50 percent of new enterprise applications are virtual machine-based and that number will reach 80 percent by 2008. Critical applications are being rolled out in financial services, telecommunications, online portals, transportation and package delivery, and retail.

Our customers, partners, analysts, and developers around the world have all made it clear that large-scale Java deployments come at a price. A few examples will suffice.

To complete the deployment of a telephone company’s Java-based billing application, the company will require hundreds of 2-way and 4-way CPU servers. Server cost isn’t an issue, but when data center managers confront significant space, power, and cooling limitations, executives wonder if there isn’t some way to consolidate servers.

An online reservations portal is scaling up its popular Java application with a number of high-end servers. Cost, however, has become an increasingly critical issue as administrators confront the real-world difficulty of predicting peak load requirements and accurately over-provisioning resources to ensure acceptable response times.

Finally, a financial services company wants to scale up its Java-based stock trading application but recognizes that its competitive advantage lies in delivering more information to users faster. With Java applications, however, as the load increases, availability can decrease. Inefficiencies such as garbage collection and limited heap size periodically bring the application to a halt. These issues are magnified in large deployments, and the company decides it cannot scale the application as it hoped.

At Azul, we believe we have the solution for each of these challenges and more. By creating a massive pool of shared compute resources, network attached processing reduces the number of required servers and the associated facilities costs. The shared resource also eliminates capacity planning at the application level, reducing wasteful underutilization and eliminating the labor cost for this complex process. And at the heart of the Azul compute appliance is our custom, multi-core chip design that provides virtual machine-based applications with all the CPU power needed while providing heap sizes up to 90GB with no garbage collection pauses, and other advantages that finally enable interpreted applications to run at speeds comparable to their chip-specific brethren.

The result is that existing Java deployments can easily deliver dramatically improved performance and cost efficiencies simply by dropping in an Azul compute appliance. At the same time, developers are free to create the next generation of more robust and innovative virtual machine-based applications, freed not only from the overhead restrictions of the interpreted language, but also from the limited number of CPUs that have restricted every application server.

The virtual machine-based platform is a great technology, and Azul will ensure that it will always work, even for the most demanding applications. But this article only scratches the surface of the issue our customers face and how network attached processing solves them. In the coming months, I will address many more such issues and lay out in even greater detail the real-world benefits of the Azul compute appliance.

Copyright © 2005, Azul Systems, Inc. All rights reserved. Azul Systems, Azul, and the Azul arch logo are trademarks of Azul Systems, Inc. in the United States and other countries. Java and all Java based trademarks and logos are trademarks or registered trademarks of Sun Microsystems, Inc, in the United States and in other countries, or both. Other marks are the property of their respective owners and are used here only for identification purposes.